Comp decisions made at 11 PM aren't decisions. They're concessions.
Every offer renegotiated at the founder's kitchen table costs more than the difference. The hidden cost is the precedent.
Here's a scene we keep walking into. The offer goes out at 100. The candidate counters at 115. It's 9 PM on a Thursday. The founder reads the email, thinks for an hour, calls the recruiter, says fine, give them 115, we need them. Done. The candidate accepts.
Two weeks later, the same candidate's eventual direct report — a person already on the team at 95 — finds out about the offer through the usual channels. They show up in the founder's office on Monday asking for a re-leveling.
The 15-point concession at 11 PM didn't cost the company 15 points. It cost 15 points times every existing employee who finds out, times the cost of the conversation they now have to have with their own manager.
What a band actually does
A band is a pre-decided answer to the 11 PM question. When the candidate counters, the founder doesn't have to think. The answer is already on a page. This is what the role pays. Here is the leveling logic. Here's the bonus structure. The founder either says yes to the band or breaks the band — and if they break it, they know they're breaking it, on purpose, for a specific reason they can defend in writing on Monday.
Most founders don't have bands because they think bands are HR bureaucracy. They aren't. They're decision insurance. They make 95% of the comp questions answer themselves before they reach the founder's inbox at 11 PM.
The smaller, harder version
You don't need a full leveling system to start. A small company can run on a four-page document. Ranges per role family. The logic for where in the range a hire lands. The policy on counter-offers. The policy on internal equity adjustments. That's enough to absorb most of the inbound.
Build that document before the next senior hire, not after.